[1] 14w Targeting demyelination via α-secretases promoting sAPPα release to enhance remyelination in central nervous system

[1] 298w Remyelination is an endogenous regenerative process of myelin repair in the central nervous system (CNS) with limited efficacy in demyelinating disorders. As strategies enhancing endogenous remyelination become a therapeutic challenge, we have focused our study on α-secretase-induced sAPPα release, a soluble endogenous protein with neuroprotective and neurotrophic properties. However, the role of sAPPα in remyelination is not known. Therefore, we investigated the remyelination potential of α-secretase-induced sAPPα release following CNS demyelination in mice. Acute demyelination was induced by feeding mice with cuprizone (CPZ) for 5 weeks. To test the protective effect and the remyelination potential of etazolate, we designed two treatment protocols. Etazolate was administrated either during the last two weeks or at the end of the CPZ intoxication. In both protocols, etazolate restored the number of myelinated axons in corpus callosum with a corresponding increase in the amount of MBP, one of the major myelin proteins in the brain. We also performed ex vivo studies to decipher etazolate's mechanism of action in a lysolecithin-induced demyelination model using organotypic culture of cerebellar slices. Etazolate treatment was able to i) enhance the release of sAPPα in the culture media of demyelinated slices, ii) protect myelinated axons from demyelination, iii) increase the number of mature oligodendrocytes, iv) promote the reappearance of the paired Caspr + adjacent to the nodes of Ranvier and v) increase the percentage of myelinated axons with short internodes, an indicator of remyelination. Etazolate failed to promote all the aforementioned effects in the presence of GI254023X, an α-secretase inhibitor. Moreover, the protective effects of etazolate in demyelinated slices were mimicked by sAPPα treatment in a dose-dependent manner. In conclusion, etazolate-induced sAPPα release protects myelinated axons from demyelination while also promoting remyelination. This work, thus, highlights the therapeutic potential of strategies that enhance sAPPα release in demyelinating disorders.

[1] 250w Demyelination in the central nervous system (CNS) is characterized by myelin sheath loss and oligodendrocyte cell death. It contributes to progressive axonal injury leading to neurological disability as observed in demyelinating diseases or in acute CNS injuries such as multiple sclerosis (MS) (Compston and Coles, 2008;Trapp and Nave, 2008) and traumatic brain injury (TBI) (Armstrong et al., 2015), respectively. In contrast, remyelination is an endogenous spontaneous regenerative process occurring in response to demyelination and is essential for the restoration of saltatory conduction and subsequent prevention of further axonal damage (Franklin and Ffrench-Constant, 2008). However, remyelination is limited or inadequate in demyelinating diseases leading to progressive axonal loss and irreversible neurological dysfunction in long-term (Dutta and Trapp, 2011). Oligodendrocyte precursor cells (OPCs), the main source of mature myelinating cells in the adult brain are not considered to be a limiting factor for remyelination since they are abundant within demyelinated regions (Chang et al., 2002). In fact, the main cause of failure in remyelination has been attributed to the inability of OPCs to differentiate into mature myelinating oligodendrocytes (Dulamea, 2017;Fancy et al., 2010;Franklin, 2002). Therefore, therapeutic strategies promoting OPCs differentiation, myelin sheath protection and remyelination may contribute to better axonal preservation and long-term functional recovery (Franklin, 2015;Kremer et al., 2015). It is noteworthy that pharmacological therapies enhancing remyelination in demyelinating diseases such as MS are still to be explored and exploited (Gajofatto and Benedetti, 2015). Therefore, our interest is focused on such strategies enhancing endogenous remyelination to promote myelin repair following demyelination.

[2] 64w In our previous report, we have shown that enhancing the release of the endogenous soluble protein sAPPα, has a therapeutic potential resulting in histological and functional improvements following TBI in mice (Siopi et al., 2013). sAPPα is released by α-secretase processing of β amyloid precursor protein (βAPP) and is known to be neurotrophic and neuroprotective (Chasseigneaux et al., 2011;Hefter and Draguhn, 2017;Mockett et al.,

[3] 116w For EM analysis, caudal part of the brain hemispheres is removed at the given coordinates (-2.0 to -3.2 mm from Bregma, Paxinos and Franklin, 2008) and then post-fixed in 4% PFA, 2.5% glutaraldehyde and 0.1M phosphate buffer, pH 7.4 for 1 week. On the day of the impregnation, caudal-lateral brain slices (-2.0 to -3.2 mm from Bregma, and 3.25 to 3.60 mm lateral) were prepared and post-fixed in 2% osmium tetroxide, dehydrated in graded ethanol series and embedded in epoxy resin. We chose to analyze the caudal-lateral region since it is one of the most demyelinated area of the corpus callosum after CPZ intake (Irvine and Blakemore, 2006;Steelman et al., 2012;Stidworthy et al., 2003;personal unpublished data).

[4] 187w Semi-thin cross-sections of the epoxy resin bloc were obtained with a glass knife at 0.5-1 μm and stained with methylene blue/azur II for quality control of the appropriate area (crosssections of lateral corpus callosum) analyzed across different experimental groups. The same resin bloc was further used to obtain ultrathin cross-sections of the lateral corpus callosum for electron microscopy analysis. Ultrathin cross-sections (50-90 nm) were performed at the level of the lateral part of the corpus callosum on an ultramicrotome (8800 Ultratome III, LKB Bromma) and collected on 300-mesh nickel grids. Staining was performed on drops of 4% aqueous uranyl acetate, followed by Reynolds's lead citrate (Reynolds, 1963). Ultrastructural analyses were performed in a JEOL jem-1011 electron microscope and digitalized with DigitalMicrograph software. Electron microscopy images were used for calculating the g-ratio (ratio of the axon diameter to the fiber diameter) and the percentage of myelinated axons using ImageJ (10.2, NIH, USA). Three animals per group and an average of ten images (magnification G x 20000) per animal were analyzed. A total number of at least 2000 axons were counted per group (at least 600 axons per animal).

[5] 158w A C C E P T E D M A N U S C R I P T 9 2.4. Western Blot Samples were taken form either brain hemispheres, cerebella and the culture media from organotypic slices were used for protein analysis. Tissue samples were homogenized in RIPA cocktail containing Tris-HCl 50 mM (pH 7.5), NaCl 150 mM, SDS 0.1% (sodium dodecyl sulfate), NP40 1% (nonyl phenoxypolyethoxylethanol), sodium deoxycholate 0.5%, sodium pyrophosphate 10 mM, EDTA 5 mM, phosphatase inhibitor (Phosphatase Inhibitor Cocktail 2, P5726, Sigma), protease inhibitor cocktail (cOmplete, Mini, EDTA-free, Sigma) and centrifuged at 12 000 x g for 15 min at 4°C. Protein concentrations were determined using a Bio-Rad DC TM protein assay kit with BSA as the standard. Three μg of protein was separated on 12% SDS-Polyacrylamide gels by electrophoresis and blotted onto polyvinylidene difluoride (PVDF) membranes (BioRad). After blocking with 5% BSA solution in Tris-Buffered Saline, membranes were probed with the following antibodies directed

[6] 65w against MBP (mouse monoclonal, 1:1000, Millipore), sAPPα (mouse monoclonal, 1:1000, BioLegend) or alpha-tubulin (mouse monoclonal, 1:10000, Sigma). They were then incubated with the appropriate horseradish peroxidase-conjugated secondary IgG antibodies (goat anti-rabbit monoclonal, 1:20 000, Jackson and goat anti-mouse polyclonal, 1:20 000, Millipore), followed by Pierce TM ECL or ECL Plus Western blotting Substrat (Thermo Scientific) before exposure to a radiographic Amersham hyperfilm ECL (GE Healthcare).

[7] 13w Relative protein amounts were quantified with ImageJ 1.48v software and normalized to alpha-tubulin.

[8] 198w All samples were washed with PBS, permeabilized for 10 min with a PBS solution containing triton-X (0.25%), gelatin (0.2%) and sodium azide (0.1%) (PBS-GTA) and then blocked with L-lysine (0.1M) in PBS-GTA for 1h. Organotypic slices were then incubated with primary antibodies overnight while samples from primary cells were incubated for 2h in a PBS-GTA The number of axonal fibers was quantified as previously described (Hussain et al., 2011;Meffre et al., 2015a and b). Quantification of axonal fibers was performed by counting the Calbindin + segments using ImageJ software (1.46r, NIH, USA). For this purpose, a straight line was traced in the region of interest localized in the apical ends of cerebellar lobules. The number of Calbindin + axons crossing this line was counted. Then, the same counting was performed for the intact myelin segment (MAG + fibers). The percentage of myelinated axons was calculated as the ratio of myelinated axons number over total axons. Three different from one category over total GFP + cells per field (0.26 mm 2 ). To determine cell proliferation, GFP + and Ki67 + cells were also quantified using ImageJ software (1.46r, NIH, USA). All quantifications have been performed by blind analysis.

[1] 32w The efficiency of etazolate on myelin protection in vivo was evaluated using a model of acute demyelination in mice treated with CPZ for 5 weeks. CPZ intoxication over a period of 5-6

[2] 123w weeks or 12 weeks is often used as a model of MS to study de(re)myelination (Kipp et al., 2009;van der Star et al., 2012). CPZ leads to oligodendrocyte loss and consequent demyelination especially in the corpus callosum and related tracks (Binder et al., 2008;Stidworthy et al., 2003;Wu et al., 2008;Yang et al., 2009). In the first protocol (n° 1), mice were treated with either vehicle or etazolate during the last 2 weeks of CPZ intoxication in order to assess the protective effect of the compound. To assess the potential of etazolate in promoting remyelination, a second protocol (n° 2) had been set up in which mice were treated with etazolate for 2 weeks at the end of the 5 weeks intoxication (CPZ withdrawal).

[3] 14w The number of myelinated axons and the g-ratio (axon perimeter / myelin fiber perimeter)

[4] 174w were determined in both protocols by electron microscopy of the corpus callosum. In protocol n° 1 (Fig. 1A), CPZ intoxication drastically reduced the percentage of myelinated axons (7% for CPZ+Veh vs 40% for Cont+Veh; P<0.0001), whereas treatment with etazolate was able to counteract this reduction (17.5% for CPZ+Etaz; P<0.01) (Fig. 1B-C). The g-ratio was also affected as it was decreased by CPZ intoxication (0.74 for CPZ+Veh vs 0.78 for Cont+Veh; P<0.0001). Similarly, etazolate treatment markedly counteracted this reduction (0.77 for CPZ+Etaz; P<0.001) (Fig. 1D). Moreover, the scatter plot displaying g-ratios of individual myelinated axons as a function of the respective axon diameter (Fig. 1E) revealed that CPZ preferentially affected smaller axons, whereas values for etazolate-treated mice were closer to the control group. In the second protocol, CPZ's effects were similar, as the percentage of myelinated axons (32% for CPZ+Veh vs 48% for Cont+Veh; P<0.05) and the g-ratio (0.72 for CPZ+Veh vs 0.75 for Cont+Veh P<0.0001) were decreased (Fig. 2A-D). Etazolate treatment could restore both of these parameters close to control values (P<0.001 and P<0.0001

[5] 42w respectively). Similarly, etazolate treatment restored the decrease in the values of g-ratio relative to smaller axons observed in the CPZ group (Fig. 2E). Finally, neither CPZ nor etazolate affected the total number of axons across our different experimental groups (data not shown).

[6] 65w Additionally, we studied the effects of etazolate treatment in vivo on the amount of MBP, one of the major proteins of central myelin. In protocol 1 (Fig. 3A), CPZ reduced the amount of MBP by 25% in the brain (75% for CPZ+Veh vs 100% for Cont+Veh, P<0.01, Fig. 3B) and by 42% in the cerebellum (58% for CPZ+Veh vs 100% for Cont+Veh, P<0.01, Fig. 3C).

[7] 204w Etazolate treatment restored MBP levels during demyelination in the brain (160% for CPZ+Etaz vs 75% for CPZ+Veh; P<0.01), as well as in the cerebellum (137% for CPZ+Etaz vs 58% for CPZ+Veh; P<0.05). Unexpectedly, we observed a positive effect of etazolate on MBP protein amount in control mice in both regions compared to vehicle-treated control mice (Fig. 3B-C, P<0.01). This observation was only limited to protocol 1 (Fig. 3B-C, 3E-F), for which the underlying mechanism has to be identified. In control groups of both protocols, etazolate treatment did not affect myelin sheath integrity, since the g-ratio remained unchanged compared to vehicle-treated control mice (data not shown). In protocol 2 (Fig. 3D) neither CPZ nor etazolate has affected MBP protein levels in the brain (Fig. 3E). The absence of CPZ effect may be due to the difference between regions analyzed for western blot compared to those used for electron microscopy. In fact, CPZ-induced demyelination can be anatomically variable (Gudi et al., 2009;Steelman et al., 2012;Stidworthy et al., 2003). In contrast, the levels of MBP were decreased in the cerebellum after CPZ intoxication (65% for CPZ+Veh vs 100% for Cont+Veh; P<0.01) and were restored by etazolate treatment (109% for CPZ+Etaz vs 65% for CPZ+Veh; P<0.01) (Fig. 3F).

[8] 14w 3.2. Etazolate protects myelin sheaths from lysolecithin-induced demyelination ex vivo: role of α-secretase activity

[9] 128w Treatment with lysolecithin induced demyelination in cerebellar slices, characterized by a disorganized MAG-staining (Fig. 4A) and a marked decrease in the percentage of myelinated axons (-42%; P<0.001) (Fig. 4B). Etazolate at 0.2 µM or 2 µM was able to counteract the myelin sheath loss (Fig. 4A), increasing the percentage of myelinated axons by 25% and 27%, respectively, compared to vehicle-treated demyelinated slices (P<0.001) (Fig. 4B). However, etazolate did not affect the myelin staining in non-demyelinated slices, since the percentage of myelinated axons was not altered as compared to control slices (Fig. 4B, N.S., P>0.999). Finally, the number of Purkinje axons (Calbindin + ) was not affected by lysolecithin or etazolate when compared to vehicle-treated slices (Fig. 4A, 4C, N.S., P=0.1859 for Veh+Lyso and N.S., P>0.9999 for other conditions).

[10] 78w In order to decipher etazolate's mechanism of action, we targeted α-secretase activity by a pharmacological inhibitor, GI254023X. When GI254023X was added together with etazolate at 2 µM, the protective effect of etazolate was no longer significant (Fig. 5A-B, N.S., P=0.4142). GI254023X when used alone did not have any deleterious or beneficial effect on myelin sheaths in demyelinated slices (Fig. 5A-B, N.S., P>0.999). In non-demyelinated slices, etazolate or GI254023X did not modify the percentage of myelinated axons (Fig. 5C).

[11] 25w Moreover, the number of Purkinje axons (Calbindin + ) was not affected by any treatment in demyelinated slices when compared to control slices (Fig. 5D).

[12] 59w these clusters was decreased in vehicle-treated demyelinated slices (P<0.001) (Fig. 6B). In contrast, etazolate treatment at 2 µM led to a reappearance of Caspr staining and thus an increase in the number of paired-Caspr + (P<0.001) (32.65 ± 3.45 for control; 1.54 ± 0.46 for lysolecithin + vehicle; 21.13 ± 2.7 for lysolecithin + etazolate) (Fig. 6A-B). When GI254023X

[13] 36w was added together with etazolate, the number of paired-Caspr + was significantly decreased compared to etazolate-treated demyelinated slices (P<0.001) (7.03 ± 1.80 for lysolecithin + etazolate + GI254023X) (Fig. 6A-B), counteracting the effect of etazolate treatment.

[14] 27w However, GI254023X alone did not modify the number of paired-Caspr + in demyelinated slices compared to vehicle-treated demyelinated slices (3.09 ± 0.93 lysolecithin + GI254023X) (Fig. 6A-B).

[15] 37w In addition, the reappearance of Caspr staining following etazolate treatment was accompanied by MAG + internode staining. Demyelinated slices treated with etazolate exhibited a high frequency of short internodes under 20 µm (P<0.001) (36% for control, 79%

[16] 78w for lysolecithin + etazolate) (Fig. 6C). Consequently, a reduction in the frequency of large internodes over 20 µm in etazolate-treated demyelinated slices was observed (P<0.001) (64% for control, 21% for lysolecithin + etazolate) (Fig. 6A, 6C). with no significant effect on the number of PDGFRα + (Fig. 7A-B). The effect of etazolate on proliferation has been also explored in vitro using PLP-GFP mice (in supplemental, Fig. S2C). While etazolate enhanced the morphological maturation of GFP + -oligodendrocytes, it

[17] 70w As our data indicated that α-secretase activation was involved in the beneficial effect of etazolate, we investigated the effect of etazolate on sAPPα release using the ex vivo model of demyelination in organotypic slices. We showed that etazolate had no significant effect on sAPPα release in control condition, whereas we observed a slight but significant increase of sAPPα levels in etazolate-treated demyelinated slices (Fig. 8, 115% for Lyso+Etaz vs 100%

[18] 51w for Lyso+Veh, P<0.05). Interestingly, our results also showed that GI254023X, the αsecretase inhibitor, drastically reduced sAPPα levels in control and lysolecithin conditions (24% for GI254023X vs 100% for Veh, P<0.05). Moreover, GI254023X abolished the sAPPα release induced by etazolate in demyelinated slices (24% for Lyso+Etaz+GI254023X vs 115% for Lyso+Etaz vs, P<0.05).

[1] 32w Our present study has brought new evidence elucidating the therapeutic potential of etazolate-induced sAPPα release in white matter protection and repair in the context of CNS demyelination ex vivo and in vivo.

[2] 111w Firstly, we investigated the protective effect of etazolate from demyelination and its ability to enhance endogenous mechanisms of remyelination in vivo. We chose a model of acute demyelination induced by cuprizone as the highly-myelinated structures like corpus callosum in the brain or the cerebellum, are affected after 5-6 weeks of CPZ intake (Kipp et al., 2009;van der Star et al., 2012). Besides, CPZ administration in mice provides for a reproducible model of spontaneous remyelination (Kipp et al., 2009;van der Star et al., 2012). Thus, two different protocols had been designed to investigate the protective effect of etazolate from demyelination (protocol 1) and its remyelinating activity after CPZ intake period (protocol 2).

[3] 13w As region-specific differences in the demyelination of the corpus callosum have been shown

[4] 539w after CPZ intake in mice, we examined the myelin sheaths by electron microscopy in the caudal region, the most demyelinated area of the corpus callosum (Binder et al., 2008;Stidworthy et al., 2003;Wu et al., 2008;Yang et al., 2009). In both protocols, we observed a reduction in the percentage of myelinated axons after five weeks of CPZ intake; in line with what has been previously reported in literature (Skripuletz et al., 2011). Interestingly, etazolate was able to partially counteract (protocol 1) and restore (protocol 2) the loss of myelinated axons due to CPZ intake. In both protocols, we also observed a reduction in the g-ratio values in demyelinated mice as previously described (Xiu et al., 2016). The g-ratio, a myelin thickness indicator, has been widely used as a functional structural index of optimal axonal myelination. CPZ's effect on g-ratio could be a consequence of the loss of thinner myelin sheaths, which leaves mainly axons with thicker myelin sheaths (i.e. low g-ratio values), leading to an overall decrease in g-ratio. Interestingly, etazolate treatment was able to restore the g-ratio values to control levels in both protocols. Thus, etazolate would mainly act on the thin myelin sheaths thereby leading to an increase in the g-ratio. As one of the main hallmarks of remyelination is an increase in the number of thin myelin sheaths (Franklin and Ffrench-Constant, 2008), we have emphasized on the remyelinating effect of etazolate (protocol 2) as well as the ability to partially protect myelin sheaths from CPZ-induced demyelination (protocol 1). We then studied the effect of CPZ and etazolate on MBP levels, one of the most abundant myelin proteins in the CNS. Moreover, it has been shown that among all the myelin proteins, MBP and CNPase were the first to be strongly degraded after 4-4.5 weeks of CPZ intake (Buss and Schwab, 2003;Skripuletz et al., 2011). We evaluated MBP protein levels in both brain and cerebellum by western blot. While there are multiple isoforms of MBP derived from differential splicing of the MBP gene (Boggs, 2006;Harauz et al., 2004), only four of them are, for now, reported at the protein level in rodents (Akiyama et al., 2002;Boggs et al., 2000;Ottens et al., 2008). We focused our analysis on the 17kDa isoform, as it is one of the forms containing exon 2 which is linked to myelination and remyelination processes (Akiyama et al., 2002;Boggs et al., 2000;Kruger et al., 1999). We showed a strong reduction in MBP levels in both protocols and regions in demyelinated mice, except in the brain in protocol 2. The lack of CPZ effect, here, is most likely due to the time period (2 weeks) following CPZ withdrawal and the beginning of the endogenous remyelination process, occurring around week 5 (Skripuletz et al., 2011). However, the effect of CPZ or CPZ+etazolate on MBP in the cerebellum was robust compared to the brain (protocol 2) probably due to the high amount of myelin in cerebellum. The beneficial effect of etazolate is probably due to its effect on enhancing the remyelination process in the cerebellum (protocol 2). Overall, our results showed that etazolate treatment was able to restore MBP protein level after CPZ-induced demyelination, indicating a beneficial effect of the compound on myelin loss and remyelination in vivo.

[5] 79w We next performed ex vivo studies to decipher the mechanism of action of etazolate. Indeed, the organotypic culture of cerebellar slices presents several advantages. It mimics the in vivo state of the cerebellar tissue, and maintains the architectural connexions, allowing an easy and fast access to the tissue, particularly useful to study myelin sheaths (Notterpek et al., 1993). Therefore, treatment of cerebellar slices with lysolecithin allows the study of demyelination and the screening of pharmacological compounds with potential remyelinating

[6] 187w effects (Zhang et al., 2011). Consistent with previous studies, lysolecithin-induced demyelination in organotypic cerebellar slice cultures led to a loss of myelin staining (MAG) and a reduction in the percentage of myelinated axons 72h following demyelination (Meffre et al., 2015a and b). Treatment with etazolate at 0.2 and 2 µM, immediately after lysolecithin removal preserves myelin sheath immunostaining leading to increased percentage of the myelinated axons, highlighting the protective effect of etazolate on myelin sheaths. In addition, a 72h-delayed treatment with etazolate increased the percentage of myelinated axons after demyelination (Fig. S1A), highlighting its remyelinating effect. As an indicator of remyelination, appearance of short myelin internodes and re-aggregation of paired-Caspr + were observed when slices were treated with etazolate immediately or 72h after lysolecithin removal (Fig. S1B). It has been shown that in demyelinated lesions, paranodal proteins such as Caspr are diffusely distributed along the axons and they no longer present a pairedaggregate distribution (Podbielska et al., 2013). However, in case of remyelination, marked changes are observed in myelin architecture as a reduction of myelin internode length and re-aggregation of paired Caspr + (Podbielska et al., 2013).

[7] 102w It is well established that the prerequisites of remyelination are OPCs recruitment and subsequent differentiation into myelinating oligodendrocytes (Franklin and Ffrench-Constant, 2008). Therefore, we studied if etazolate affects OPCs differentiation and provided evidence that etazolate treatment enhances i) oligodendrocyte differentiation ex vivo by an increase in the number of mature CC1 + oligodendrocytes and ii) morphological maturation of oligodendrocytes in vitro by an increase of multi-branched cells (Fig. S2B) without promoting cell proliferation (Fig. S2C). In addition to oligodendrocytes, etazolate could also promote remyelination by acting on neurons, astrocytes or microglia (Barnett and Linington, 2013;Domingues et al., 2016;Miron, 2017;Simons and Trajkovic, 2006).

[8] 55w Concerning the action mechanism of etazolate, it has been reported that this compound can act on multiple targets (Barnes et al., 1983;Chasin et al., 1972;Daly et al., 1988;Marcade et al., 2008;Thompson et al., 2002). In order to decipher its mechanism of action in our experimental condition, we focused our study on α-secretases promoting sAPPα release

[9] 156w A C C E P T E D M A N U S C R I P T (Marcade et al., 2008). In fact, we have shown that etazolate treatment was able to enhance sAPPα release in the culture media of demyelinated slices, which was reverted by an αsecretase inhibitor GI254023X. We have to emphasize that evaluation of etazolate-inducing sAPPα over-release in the culture media of cerebellar slices appears to be underestimated due to possible protein retention in the extra-cellular matrix, or its recapture by a receptor, still undiscovered (see Kögel et al., 2012). In fact, etazolate is known to increase dosedependently the release of sAPPα in vitro (see (Marcade et al., 2008), personal unpublished data). Interestingly, the protective effect of etazolate on myelin sheaths and its remyelinating effect were attenuated or even blocked in the presence GI254023X. It is noteworthy that a marked decrease in sAPPα release by GI254023X was not deleterious to myelinated axons.

[10] 246w In our experimental setting, the cerebellar slices reach the maximum of myelination before starting any treatment protocols (Birgbauer et al., 2004;Dusart et al., 1997;Ghoumari et al., 2002). The 3-days treatment of slices with GI254023X did not have any deleterious effect (demyelinating) in our cerebellar slices since these slices are already fully myelinated. The fact that GI254023X, decreasing sAPPα release, did not have any effect on myelinated axons is precisely relevant. However, further investigations are required to clarify the role of sAPPα on the maintenance of myelin sheath integrity. Overall, our data highlight that αsecretases are involved in the beneficial effects of etazolate on oligodendrocyte differentiation/maturation and on remyelination. GI254023X, as α-secretase inhibitor, is more specific to ADAM10 (IC 50 5 nM) than ADAM17 (IC 50 541 nM) and is able to inhibit other metalloproteinases such as MMP9 (IC 50 2.5 nM) (Ludwig et al., 2005). Therefore the role of these enzymes in the beneficial effects of etazolate should be studied further. While ADAM17 expression is regulated (Arribas and Esselens, 2009), ADAM10 is constitutively expressed in the brain (Kärkkäinen et al., 2000), cerebellum (Guo et al., 2016), NPCs (Klingener et al., 2014) and oligodendrocytes (Lin et al., 2008). Even though the role of αsecretases in CNS remyelination is not fully understood (Klingener et al., 2014;Palazuelos et al., 2015Palazuelos et al., , 2014)), both ADAM10 and ADAM17 may play a role in the remyelination potential of etazolate by enhancing sAPPα release. Indeed, sAPPα is a multifunctional protein which

[1] 70w Data are expressed as mean  standard error of the mean (s.e.m). The data were analyzed using GraphPad Prism statistical software (Prism v4, GraphPad, La Jolla, CA). One-way ANOVA followed by Bonferroni's multiple comparison post-test was used for the comparison of three or more groups. For Western blot analysis, non-parametric Kruskal-Wallis followed by U-Mann-Whitney test was used. Differences with a P value under 0.05 were considered to be statistically different.

[1] 76w M A N U S C R I P T 5 2017). However its effect on white matter protection or repair remains unknown. The αsecretases include different members of ADAM (A Disintegrin And Metalloproteinase) family such as ADAM9, ADAM10 and ADAM17 (Allinson et al., 2003). Within the brain, ADAM10 is the most relevant α-secretase since it leads to a constitutive and regulated α-secretase cleavage of βAPP under physiological conditions (Kuhn et al., 2010;Prox et al., 2013).

[2] 61w Although the role of α-secretases in remyelination has not been fully investigated, it has been shown that i) neuronal ADAM10 activity is required for the migration of neural precursor cells (NPC) into demyelinated lesions (Klingener et al., 2014), and that ii) the modulation of oligodendroglial ADAM17 promotes oligodendrogenesis and myelin repair in vivo (Palazuelos et al., 2015(Palazuelos et al., , 2014)).

[3] 77w Therefore, the purpose of this study was to investigate the protective and remyelination potential of a pharmacological strategy able to promote sAPPα release following demyelination. Etazolate, a pyrazolopyridine derivative, is known to enhance the α-secretase processing of βAPP to sAPPα and also promotes neuroprotection from Aβ peptide neurotoxicity in vitro (Marcade et al., 2008) and post-TBI complications in vivo (Siopi et al., 2013). Nevertheless, the therapeutic potential of etazolate in myelin sheath protection and remyelination remains unknown.

[4] 139w Hence, we investigated the protective and remyelinating effects of etazolate as an αsecretase activator in C57BL/6 mice. We studied the effect of etazolate treatment at the dosage known to enhance sAPPα release in vivo (Marcade et al., 2008) in the cuprizone (CPZ)-induced acute demyelination model. We designed two treatment protocols to test i) the protective effect and ii) the remyelination potential of etazolate. We also used the model of lysolecithin-induced demyelination in organoytpic culture of cellebellar slices, a well known ex vivo model to study de(re)myelination processes. The latter was used to study the mechanism of action of etazolate, at the dosage known to increase sAPPα release (Marcade et al., 2008) and also in the presence of GI254023X, an α-secretase inhibitor (Jangouk et al., 2009;Ludwig et al., 2005). Our results showed for the first time that etazolate promotes white

[5] 175w Eight-week-old C57BL/6 male mice (n=50/protocol) were randomly assigned to different experimental groups after 1 week of habituation. Acute demyelination was induced by feeding mice with 0.2% (w/w) cuprizone (bis(cyclohexanone)oxaldihydrazone) (Sigma-Aldrich) mixed in rodent chow for 5 weeks (Skripuletz et al., 2011). Control group received a normal diet without cuprizone. After 3 weeks of cuprizone intoxication (Protocol n°1), animals were treated intraperitoneally (i.p.) for 2 weeks either with a vehicle (PBS) or etazolate (10 mg/kg/day), a dose known to increase sAPPα levels in vivo (Marcade et al., 2008) (n=10 per group). In the second protocol (Protocol n°2), mice were intoxicated with cuprizone for 5 weeks and then allowed to recover for 2 weeks by feeding on normal chow. During this period, animals were treated either with a vehicle (PBS) or etazolate (10 mg/kg/day, i.p.). For histological studies, mice were deeply anesthetized with pentobarbital (60 mg/kg, i.p.), and perfused through the heart with phosphate buffer 0.1M (pH 7.4) before brain removal. The caudal part of brain hemispheres were immersed in a cold fixative solution (4% PFA, 2.5%

[6] 57w A C C E P T E D M A N U S C R I P T 8 glutaraldehyde and 0.1M phosphate buffer, pH 7.4) and stored at 4°C until use. For biochemical studies, mice were sacrificed by cervical dislocation, and the brains were removed, instantly frozen in liquid nitrogen and stored at -80°C until use.

[7] 80w Organotypic cerebellar slices were obtained from C57BL/6 mouse pups aged between 8 to 10 post-natal days (P8-P10), as described previously (Hussain et al., 2011;Meffre et al., 2015a and b) . Brains were removed and placed in cold-PBS supplemented with 5 mg/mL Dglucose (Sigma-Aldrich). After removal of the meninges, the cerebellum was dissected and 350 m-thick parasagittal slices were cut using a tissue chopper (McIlwain Tissue Chopper, UK). Isolated slices were transferred onto membranes of 30 mm Millipore culture inserts with

[8] 101w A C C E P T E D M A N U S C R I P T 10 0.4 m pore size (Millipore, Bedford, MA). The inserts were placed in 6-well tissue culture plates with 1 mL of medium containing 50% basal medium with Earle's salts (Invitrogen, Gaithersburg, MD), 2.5% Hank's balanced salt solution (Life Technologies, Grand Island, NY), 25% heat-inactivated horse serum (Life Technologies), glutaMAX® 1% (Life Technologies), 5 mg/mL D-glucose (Sigma-Aldrich) and penicillin (0.1 U/mL)-streptomycin (0.1 µg/mL). Slices were maintained at 35°C in an atmosphere of humidified 5% CO 2 and the medium was changed every 3-4 days.

[9] 23w 2.6. Lysolecithin-induced demyelination in organotypic cerebellar slice cultures At 7 DIV (days in vitro), slices were treated with lysolecithin (0.5 mg/mL) (PubChem CID:

[10] 123w 24798682; Sigma-Aldrich) or its vehicle (0.5% methanol:chloroform, 1:1) for 16h, as previously described (Birgbauer et al., 2004;Meffre et al., 2015a and b). The slices were then treated with a vehicle (PBS 0.1M and DMSO 0.025% mixture) or etazolate (PubChem CID: 3277; Tocris Bioscience) at the dose of 0.2 or 2 µM to promote sAPPα release (Marcade et al., 2008), or etazolate (2 µM) with GI254023X (PubChem CID: 9952396; Tocris Bioscience) at 2 µM, to block α-secretase activity (Jangouk et al., 2009). After 72h of treatment, cerebellar slices were fixed with 4% paraformaldehyde solution (PFA) for subsequent immunolabelling. In another set of experiments, we studied the effect of sAPPα at the dose of 5 and 20 nM (recombinant human sAPPα, Sigma) on demyelinated slices.

[11] 92w Since no study has addressed the effect of sAPPα on de(re)myelination, we used the dosing range of sAPPα at a nanomolar concentration that has been previously shown to be neurotrophic in vitro (Billnitzer et al., 2013;Chasseigneaux et al., 2011) 2.7. Primary mixed glial culture Newborn (P1-P3) PLP-eGFP mouse pups were decapitated and the hemispheres were collected in cold-PBS containing 5 mg/mL glucose. After removal of the meninges, the hemispheres were mechanically dissociated into DMEM (Invitrogen, France) and the solution obtained was filtered (100 µm, BD Biosciences) and seeded into 12-well plates containing

[12] 117w one glass cover-slide per well (Ø 14 mm, Knittel Glass, Germany), previously coated by 20 µg/mL of poly-L-lysine (Sigma, St Louis, Missouri, USA). Cells were incubated at 37°C in an atmosphere of humidified 5% CO 2 , in a medium containing 10% Fetal Bovine Serum (Invitrogen, France), L-glutamine 2 mM, sodium pyruvate 1 mM, fungizone 0.5 µg/mL, penicillin (0.1 U/mL)/streptomycin (0.1 µg/mL) diluted in DMEM. Medium was changed at 5 DIV and then every 2 days. Cells were treated at 7 DIV, either with vehicles (PBS and DMSO 0.025% mixture), etazolate at 2 µM, GI254023X at 2 µM, or a combination of both compounds. After 72h of treatment, cells were fixed with 4% PFA for subsequent immunolabelling.

[13] 13w M A N U S C R I P T 14 3. Results

[14] 42w It is well known that remyelination induces changes in myelin architecture such as i) reaggregation of Caspr proteins expressed in the paranodal junctions, adjacent to the nodes of Ranvier, and ii) reduction in the length of remyelinated internodes (Podbielska et al., 2013).

[15] 23w Therefore, the remyelination potential of etazolate was investigated by labeling Caspr clusters and MAG in order to highlight paranodal junctions and remyelinated internodes.

[16] 24w Lysolecithin-induced demyelination led to a diffused Caspr staining with no further aggregation of paired-Caspr + 72h post demyelination (Fig. 6A). Therefore, the number of

[17] 42w did not alter the number of Ki67 + /GFP + proliferating cells (Fig. S2B-C). However, when GI254023X was added together with etazolate, the number of CC1 + cells was significantly reduced by 26% compared to etazolate-treated demyelinated slices (P<0.01) (Fig. 7A, 7C).

[18] 48w Treatment with GI254023X alone did not exert a significant effect on the number of CC1 + or PDGFRα + cells, compared to vehicle-treated demyelinated slices (Fig. 7A-C). Furthermore, neither etazolate nor GI254023X changed the number of PDGFRα + and CC1 + cells in nondemyelinated slices (data not shown).

[19] 73w Since the mechanism of action of etazolate was related to sAPPα release, we then investigated the effect of sAPPα on demyelinated slices. As shown in Figure 9, sAPPα was able to protect myelin sheaths in a dose-dependent manner. We observed an increase of 14% (P<0.05) and 34% (P<0.001) in the percentage of myelinated axons when sAPPα was added to demyelinated slices at a dose of 5 nM and 20 nM, respectively (Fig. 9A-B).

[20] 17w Treatment with sAPPα in control and demyelinated slices did not modify the number of axons (Fig. 9C).

[21] 134w promotes neuroprotection, neurotrophism, neurogenesis, synaptic plasticity and memory (Mockett et al., 2017). In addition to these beneficial effects in the CNS, our work highlights the relevance of this endogenous protein in demyelinating disorders. The mechanisms underlying this new property of sAPPα remains to be determined. In addition, α-secretases are able to cleave a variety of substrates (Edwards et al., 2008) and the role of other released products, in addition to sAPPα, in the remyelinating effect of etazolate should not be underestimated. For example, Neuregulin1 can be cleaved by ADAM10/ADAM17 (Hu et al., 2016) and the overexpression of its secreted form has been shown to enhance the expression of PLP and MBP in vivo (Luo et al., 2014). Further studies are required to determine the involvement of such molecule in the remyelinating effect of etazolate.

[22] 746w Taken together, we described the therapeutic interest of etazolate in white matter protection and repair by protecting myelin sheaths, enhancing oligodendrocyte differentiation and the remyelination process after demyelination. Our work highlights the role of α-secretases in etazolate-induced sAPPα release and myelin repair. In addition, etazolate presents a good tolerability and safety profile in humans (Vellas et al., 2011) and exerts other benefits such as anti-inflammatory (Guo et al., 2014;Siopi et al., 2013), pro-cognitive (Drott et al., 2010), antidepressant and anxiolytic (Ankur et al., 2013;Jindal et al., 2012) activities in animal models. Our work provides an additional evidence that etazolate-induced sAPPα release protects myelinated axons from demyelination and enhances remyelination, highlighting the therapeutic potential of strategies to enhance sAPPα release in demyelinating disorders. protects from demyelination in vivo. (A) Timeline of the experiment (Protocol 1). (B) Representative electron micrographs obtained from the cross-sections of the lateral corpus callosum of controls, CPZ and CPZ+etazolate-treated mice. Scale bar: 0.5 µm. Quantification of electron micrographs show a strong reduction in the percentage of myelinated axons (C) in addition to a reduction in g-ratio (D) with CPZ. Etazolate treatment protects against loss of myelinated axons and g-ratio reduction. (E) Scatter plot displaying g-ratios of individual myelinated axons as a function of the respective axon size. A total number of at least 600 axons were counted per animal. Results are expressed as mean ± of three animals per group. **P<0.01; ***P<0.001; ****P<0.0001 by ANOVA one way followed by Bonferroni's post test. Cont (Control, Normal Chow); CPZ: Cuprizone 0.2%; Etaz: Etazolate 10mg/kg/d; Veh (Vehicle, PBS). induced demyelination in vivo. (A) Timeline of the experiment (Protocol 2). (B) Representative electron micrographs obtained from the cross-sections of the lateral corpus callosum of controls, CPZ and CPZ+etazolate treated mice. Scale bar: 0.5 µm. Quantification of electron micrographs show a reduction in the percentage of myelinated axons (C) in addition to a reduction in g-ratio (D) with CPZ. Etazolate treatment enhances remyelination as it increases the percentage of myelinated axons and the g-ratio. (E) Scatter plot displaying g-ratios of individual myelinated axons as a function of the respective axon size. A total number of at least 600 axons were counted per animal. Results are expressed as mean ± demyelination in vivo. Western blot and quantification of MBP in either the brain (B, E) or the cerebellum (C, F) for the two different protocols used (A, D). In Protocol 1, cuprizone intoxication induces a reduction of MBP amounts and etazolate counteracts this effect, both in the brain and the cerebellum. In Protocol 2, there is neither a CPZ nor etazolate effect in the brain (E). In the cerebellum, CPZ decreases the MBP amounts whereas etazolate restores them. Results are expressed as mean ± s.e.m obtained from 4 independent experiments *P<0.05; **P<0.01; by Kruskal-Wallis followed by U-Mann-Whitney. Control (Control, Normal Chow); CPZ: Cuprizone 0,2%; Etaz: Etazolate 10 mg/kg/d; Vehicle (PBS). induced demyelination. (A) Confocal images of double immunostaining for calbindin (Purkinje cells, red) and MAG (myelin sheaths, green). Scale bar: 50 µm. (B) Effect of lysolecithin and etazolate at 0.2 and 2 µM on the percentage of myelinated axons. (C) Effect of pharmacological compounds on the total number of axons in control and demyelinated slices. Results are expressed as mean  s.e.m of 3 different regions (0.2 mm 2 ) quantified per stain and per slice, from 24 slices per condition obtained from 4 independent experiments. ***P<0.001 (One-way ANOVA following Bonferroni's post-test). Calbindin: calcium-binding protein; Ctl: control slices; Etaz 0.2: etazolate 0.2 µM; Etaz 2: etazolate 2 µM; MAG: myelin associated glycoprotein; Veh: vehicle (PBS). calbindin (Purkinje cells, red) and MAG (myelin sheaths, green). Scale bar: 50 µm. Effect of junctions, red) and MAG (myelin internodes, green). Scale bar: 50 µm. A 3X magnification of paired-Caspr + pointed by close arrowheads and internodes are shown in the second row of images. Scale bar: 10 µm. (B) Effect of pharmacological compounds on the number of paired-Caspr + . (C) Effect of pharmacological compounds on the frequency of short internodes (<20 µm) and long internodes (>20 µm). Because of alteration in MAG staining in lysolecithin conditions, internode lengths of demyelinated axons were not calculable. Results are expressed as mean  s.e.m of 4 to 6 different regions (0.05 mm 2 ) quantified per slice, from 12 slices per condition obtained from 2 independent experiments. ***P<0.001 (One-way ANOVA following Bonferroni's post-test). Caspr: contactin-associated protein; Ctl: control slices; Etaz: etazolate 2 µM; GI: GI254023X 2 µM; MAG: myelin associated protein; Veh: vehicle (PBS+DMSO).